Impact of Aggregate Size and Structure on the Photocatalytic

Jan 5, 2012 - Andreas P. Gondikas , Frank von der Kammer , Robert B. Reed , Stephan Wagner , James ..... P. Chandrasekaran , P. Anandan , N. Srinivasa...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/est

Impact of Aggregate Size and Structure on the Photocatalytic Properties of TiO2 and ZnO Nanoparticles David Jassby, Jeffrey Farner Budarz, and Mark Wiesner* Department of Civil and Environmental Engineering, Pratt School of Engineering, Duke University, Durham, North Carolina 27708, United States S Supporting Information *

ABSTRACT: Aggregation of photocatalytic semiconductors was determined to reduce the generation of free hydroxyl radicals in aqueous suspensions in a fashion dependent on aggregate size and structure. Static light scattering measurements were used to follow temporal changes in the fractal dimension of aggregating TiO2 and ZnO nanoparticles. At length scales comparable to nanoparticle size, the structure of aggregated TiO2 nanoparticles was independent of particle stability and the associated aggregation rate, consistent with the fused nature of TiO2 primary particles in the initial suspension. In contrast, ZnO aggregates were characterized by smaller fractal dimensions when ionic strength, and the resulting aggregation rate, were increased. The photocatalytic activity of ZnO and TiO2 in generating free hydroxyl radicals varied with aggregate structure and size, consistent with theory that predicts reduced reactivity as aggregates become larger and more dense.

■

INTRODUCTION Photocatalytic semiconducting nanoparticles are being studied for a wide range of applications including water treatment, solar power, and self-cleaning surfaces.1−5 TiO2 nanoparticles, arguably the most commonly used engineered nanomaterial, are found in numerous industrial products and applications.6−8 ZnO is widely studied for its photocatalytic ability and is used in many commercial applications. Because of the wide use of these materials, there is growing concern over their release into the environment.9−11 Once introduced to the environment, nanoparticles encounter conditions of ionic strength, pH, and solution chemistry that may alter nanoparticle surface chemistry, leading to changes in stability with respect to aggregation.12,13 The aggregation of metal oxide nanoparticles has been found to decrease photocatalytic properties in generating reactive oxygen species.14−18 However, the roles that aggregate size and structure may play in producing this decrease have not been demonstrated. While mass transfer and shadowing effects can be expected to reduce reactive oxygen species (ROS) generation when nanoparticles aggregate, the theory we have developed for surface-active nanoparticles also suggests that the very proximity of reactive surfaces may play a role in determining the net nanoparticle reactivity.19 Aggregating particles may associate in a fractal, or self-similar, structure.20,21 The mass of primary particles in a fractal aggregate plotted as a function of length scale is observed to be linear over intermediate length scales, with a noninteger value for slope that describes the fractal dimension (D) of an aggregate.22 As D © 2012 American Chemical Society

approaches a value of 3, aggregates approach their maximum density. Conditions favoring particle−particle attachment during aggregation, and therefore rapid aggregation, lead to aggregates with a lower D, compared to slower aggregation processes.22 Particle aggregation can be induced through the addition of an electrolyte such as NaCl. The added electrolyte causes a screening of the surface charge, leading to a decrease in electrostatic repulsion between two similarly charged surfaces. Once the electrostatic repulsive forces are reduced, particles can be bound together through attractive van der Waals forces.22 The primary form of ROS generated by semiconducting metal oxides is the hydroxyl radical,23 created when an electron hole (formed when the energy of absorbed photons exceeds the band gap of the material) reacts with a water molecule or hydroxyl ion on the particle surface.6,24 Processes such as aggregation affect the local concentration of one of the key reactantsparticle surfacesand may thereby alter ROS production. In this work, we consider the effect of the size and structure of aggregates of TiO2 and ZnO nanoparticles on the generation of free hydroxyl radicals. The focus of the study is on physicochemical properties of photocatalytic nanomaterials, and how aggregation impacts these properties. We find that the Special Issue: Transformations of Nanoparticles in the Environment Received: Revised: Accepted: Published: 6934

June 13, 2011 December 8, 2011 January 5, 2012 January 5, 2012 dx.doi.org/10.1021/es202009h | Environ. Sci. Technol. 2012, 46, 6934−6941

Environmental Science & Technology

Article

were added to 1 mL of stock nanoparticle suspension to make up a total volume of 2 mL. The suspension was placed in the SLS/DLS instrument and measurements were taken every 10 min. All aggregation experiments were done in triplicate. For EPM measurements, a 5 ppm nanoparticle suspension (TiO2 or ZnO) was used with varying NaCl concentrations (0−0.05 M NaCl for ZnO; 0−0.2 M NaCl for TiO2). The absorption cross section of an aggregating sample was determined using a Varian Eclipse UV−vis spectrophotometer (Agilent, Santa Clara, CA). In these experiments, 4-mL disposable plastic cuvettes loaded with 1 mL of stock nanoparticle suspension, 0.5 mL of stock TA solution, varying volumes of stock NaCl solution, and enough DIW to make up 2 mL were placed into the instrument. Absorption measurements were taken every 10 min between 340 and 390 nm. All absorption experiments were done in triplicate. Measurements of hydroxyl radical generation were done using 4-mL disposable plastic cuvettes loaded with 1 mL of stock nanoparticle suspension, 0.5 mL of stock TA solution, varying volumes of stock NaCl (same as in the SLS/DLS experiment) solution, and enough DIW (same as in SLS/DLS experiment) to make up 2 mL. In these experiments, we exposed a single cuvette to UV irradiation for 1 min, and did not expose each cuvette to UV more than once. Therefore, for each electrolyte concentration, a unique cuvette was prepared for each time point. Measurements were taken every 10 min. We initiated aggregation in all cuvettes simultaneously by adding NaCl to each one, but only exposed 1 cuvette to UV at a given time point. UV exposure was done in a UV box equipped with a low-pressure mercury fluorescent lamp (TL-D 15W BLB SLV, Philips, Eindhoven, Netherlands) with emission centered around 365 ± 25 nm. Following the UV exposure, 1.5 mL of the sample was centrifuged for 5 min at 18 000 rpm, and 1 mL of supernatant was removed. Photoluminescence (PL) of TAOH was measured using a Varian Eclipse fluorometer (Agilent, Santa Clara, CA), with excitation wavelength set at 315 nm and emission set at 425 nm. All UV exposure experiments were done in triplicate.

data are well described by theory that considers interactions between particle surfaces within an aggregate, analogous to a previously developed theory published by our group that describes the impact of aggregation on the ability of C60 to generate ROS in the form of singlet oxygen.19 The results of this study emphasize the importance of aggregate size and structure beyond simple mass transport limitations, with significance for interpreting nanomaterial behavior in natural systems as well as in engineered applications where nanoparticles may aggregate or be deposited on substrates.

■

MATERIALS AND METHODS Materials. P25 TiO2 powder was generously provided by Evonik (Evonik Industries, Essen, Germany). ZnO powder was purchased from Skyspring Nanomaterials Inc. (Houston, TX). NaCl, NaOH, K2HPO4, and KH2PO4 were purchased from Fisher Scientific (Pittsburgh, PA). Hydroxyl radical production was measured using terephthalic acid (TA, Sigma, St. Louis, MO), which has been found to be a selective free hydroxyl radical probe, producing 2-hydroxy terephthalic acid (TAOH), and has been used extensively to evaluate the ability of TiO2 to generate hydroxyl radicals.25−27 TAOH fluoresces at 432 nm, while TA does not, allowing for the quantification of free hydroxyl radicals in solution. TiO2 stock suspension was prepared by adding 40 mg of TiO2 P25 powder to 1 L of deionized water buffered to pH 7.8 using 10 mM phosphate buffer, followed by 30 min of probe sonication (20 kHz, 120 μm). ZnO stock suspension was prepared by adding 80 mg of ZnO powder to 1 L of DIW adjusted to pH 10.8 with NaOH, followed by 30 min of probe sonication. TA stock solution was made at a concentration of 0.5 mM in DIW adjusted to pH 9.8 with NaOH. NaCl stock solution (1 M) was prepared by stirring NaCl in DIW. Particle Characterization. Measurements of particle and aggregate electrophoretic mobility (EPM) were done using a Zeta Sizer Nano ZS (Malvern, Bedford, MA). A drop of TiO2 or ZnO stock suspension was placed on a lacey carbon (TiO2) or holey carbon (ZnO) grid, followed by drying at room temperature. Grid-deposited nanoparticles were visualized by a transmission electron microscope (TEM) (FEI Tecnai G2 Twin, Hillsboro, OR) operating at 160 kV. Aggregate size and structure were determined using a light scattering setup (ALV-CGS3, ALVgmbh, Langen, Germany) employing a He/ Ne laser (632.8 nm) and a detector mounted on a goniometer. Dynamic light scattering (DLS) measurements were effected with this setup to determine the average hydrodynamic radius (Rh) from time-variable light scattering at a scattering angle of 90°. Static light scattering (SLS) measurements at angles ranging from 60° to 140°, using the same setup as for DLS, were used to characterize aggregate structure. For each angle measured using SLS, three 5-s measurements were used. However, a measurement was used only if the count rate was within 5% of the previous measurements count rate. Otherwise, the measurement was repeated. Typical acquisition times for the entire spectrum were on the order of 7 min. Due to this time delay, all SLS measurements are by necessity timeaveraged. Experimental Procedure. In the SLS/DLS experiments, 0.5 mL of stock TA solution, varying volumes of stock NaCl solution (0.4, 0.1, 0.08, 0.07, and 0.04 mL to create a final NaCl concentration of 0.2, 0.05, 0.04, 0.035, and 0.02 M NaCl, respectively), and enough DIW (0.1, 0.4, 0.42, 0.43, and 0.46 mL for 0.2, 0.05, 0.04, 0.035, and 0.02 M NaCl, respectively)

■

RESULTS AND DISCUSSION TEM images of the initial particle dispersions (prior to the addition of NaCl) were obtained for the purpose of comparing starting conditions. Images of the initial TiO2 suspensions (40 ppm) revealed an absence of individual primary particles (Figure 1a). In contrast, TEM images of the initial ZnO suspensions (80 ppm) showed well-separated primary particles (Figure 1b). These images are consistent with many published observations of the TiO2 used in this study (see for example 28 and 29)

Figure 1. TEM images of TiO2 (a) and ZnO (b). 6935

dx.doi.org/10.1021/es202009h | Environ. Sci. Technol. 2012, 46, 6934−6941

Environmental Science & Technology

Article

Figure 2. Changes in the fractal dimension of aggregating TiO2 and ZnO suspensions with varying electrolyte concentrations.

differences in aggregate fractal dimension, regardless of aggregation rate (Figure 2). At 50 and 60 min, the slower aggregating TiO2 suspensions (corresponding to the lower NaCl concentrations of 0.04 and 0.035 M) exhibited slightly higher fractal dimension values than the fast aggregating particles (corresponding to the higher NaCl concentration of 0.2 M). Values of fractal dimension ranged between 1.98 and 2.18. In contrast, aggregation rate had a larger impact on ZnO aggregate fractal dimension (Figure 2). For the 0.05 M aggregation (rapid aggregation), fractal dimension values ranged between 2.07 and 2.12. However, 0.02 M aggregation (slow aggregation) yielded much denser aggregates as expressed by their significantly higher fractal dimension ranging between 2.34 and 2.40. In general, measurements in early stages of aggregation are more susceptible to interference from aggregation, and are thus less robust. To verify that changes in D were indeed a result of aggregation, and not simply an artifact resulting from multiple scattering by the particles, we aggregated 20 ppm of ZnO nanoparticles (half of the concentration used in the ROS generation experiments) using 0.02 M NaCl. The resulting fractal dimension of the aggregates was similar to that produced with 40 ppm ZnO, indicating that multiple scattering coupled with aggregation was not responsible for the observed change in structure (Figure S3). Whereas the ZnO suspension begins as a stable suspension consisting of primary particles 10−30 nm in diameter, the TiO2 suspension is made up of fused primary particles, with Rh = 125 nm. Light scattered from TiO2 aggregates is dominated by scattering from fused TiO2 primary particles. The packing of primary particles within an aggregate is not affected by further aggregation as they are fused and will not change with further aggregate−aggregate joining. Therefore, aggregation of these fused primary particles will preserve the structure of the fused aggregate at larger length-scales, resulting in a fractal dimension that exhibits only minor changes, regardless of the aggregation rate. This is consistent with previously reported results about cluster−cluster aggregate structure.30,31

showing stable aggregates of fused primary particles despite sonication. This is particularly evident along aggregate edges (Figure 1a). By trial and error, we found that a 90° scattering angle for the DLS measurements angle exhibited the most consistent results for the range of particle radii (10−1000 nm) we were investigating. The Rh of both TiO2 and ZnO dispersions showed a clear dependence on both time and electrolyte concentration (Figure S1a,b). DLS measurements of the stock TiO2 suspension indicate the presence of particles with Rh = 125 ± 31 nm. Data fits to the autocorrelation function (using ALV software package) yielded an intensityweighted hydrodynamic radius of 116 ± 5 nm, and no evidence of smaller particles in the suspension. DLS measurements of the stock ZnO suspension yielded Rh = 90 ± 18 nm, however data fits indicated two peaks at hydrodynamic radii of 99 ± 4 nm (99%) and 20 ± 9 nm (1%) when weighted by intensity. A number-weighted fit of the DLS generated autocorrelation function indicated that 75% of the particles were near the smaller 20 nm peak. As commonly observed, increasing electrolyte concentrations increased the aggregation rate. This phenomenon is consistent with measurements of EPM (Figure S1c), where increasing electrolyte concentrations caused the EPM to approach zero. Temporal changes to TiO2 and ZnO fractal dimensions were followed using SLS. The fractal dimension of the aggregates in suspension was evaluated by determining the slope of a line formed by plotting the inverse of the intensity of the scattered light (log(I)) vs the scattering vector (log(Q)). Q values were in the range of 0.013−0.024 nm−1. These Q values correspond to length scales ranging between 252 and 475 nm. Adding angles below 60° (Q value below 0.013 nm−1) did not significantly change the measured D values, but did increase the measurement time and increased noise at those angles. The length scales covered in this study include length scales common to all aggregates examined in this study. A complete set of the inverse of scattering intensities vs scattering vectors is shown in Figure S2. For aggregating TiO2 nanoparticles, the first 40 min of aggregation did not yield statistically significant 6936

dx.doi.org/10.1021/es202009h | Environ. Sci. Technol. 2012, 46, 6934−6941

Environmental Science & Technology

Article

To monitor the effect of particle aggregation on the ability of TiO2 and ZnO to generate free hydroxyl radicals, we employed a chemical probe (TA) that photoluminesces upon interaction with a free hydroxyl radical.25 TA is negatively charged at the pHs (pH 10.8 for ZnO and 7.8 for TiO2) used in this study (pKa2 = 4.8). The low concentration of TA used in this study (0.125 mM TA) was insufficient to induce aggregation by itself. The experimental design ensured that the TA transformation to its hydroxylated form was not mass limited by using high concentrations of this marker (0.125 mM) relative to the nanoparticle concentrations, exposing the suspension to UV light for only brief periods (1 min), and by not re-exposing the same suspension to UV twice. Thus, as the aggregate grew, TA was included in the aggregate pores, and due to the brief UV exposure we can assume that no TA needed to diffuse from the bulk to replenish TA concentrations within the aggregate. No pH changes were observed in either the TiO2 or ZnO suspensions after the short exposure to UV. Furthermore, the high pH (10.8) used in the ZnO suspensions minimized particle dissolution.32,33 Dissolved ionic species have a detrimental impact on the ability of photocatalytic nanoparticles to generate ROS.34−36 Therefore, comparing the impact of aggregation on the ability of TiO2 and ZnO to generate hydroxyl radicals in suspensions with different ionic strengths is not practical due to the additional quenching inherent in the suspension with higher ionic strength. Thus, in our study, the impact of aggregation on hydroxyl radical generation was limited to a given ionic strength (0.2, 0.05, 0.04, 0.035, and 0.02 M), i.e., changes in hydroxyl radical generation capacity (measured as changes in TAOH PL) of an aggregating suspension were compared to the capacity at time zero when the electrolyte was just added to the suspension, but little, if any, aggregation took place. The process of aggregation had a detrimental impact on the photocatalytic activity of TiO2 nanoparticles, decreasing hydroxyl radical generation by over 15% for large aggregates (Figure S4a). Comparing OH• generation relative to Rh yields conflicting results. In the case of TiO2 aggregates formed with 0.035 M NaCl, it appears that the slowly aggregating particles generate less OH• than rapidly aggregating particles (0.2 M) (Figure S4a). In contrast, for some Rh values, it appears that the slightly faster aggregating TiO2 (0.04 M) generate more OH• than the rapidly aggregating particles (0.2 M) (Figure S4a). To account for this discrepancy, we normalized the PL by the number of primary particles in each aggregate. Since the TiO2 aggregates exhibit mass fractal properties, the number of primary particles per aggregate can be calculated as19 ⎛ d ⎞D n=⎜ a⎟ ⎝ dm ⎠

Figure 3. Normalized (by number of primary particles in aggregate) PL vs aggregate size: (a) aggregating TiO2; (b) aggregating ZnO. Concentrations in the legend refer to NaCl concentrations.

our previous claim that TiO2 aggregate structure at a scale relevant for photocatalysis does not significantly change with changing aggregation rate. When comparing PL to Rh in the aggregating ZnO suspension (Figure S4b), it appears that the hydroxyl radical generation capacity is similar for both aggregation rates for aggregates up to 350 nm in diameter; larger aggregates formed under slower aggregation regime (0.02 M) displayed a decreased ability to generate hydroxyl radicals when compared to aggregates formed in the fast aggregation regime (0.05 M; Figure S4b). However, comparing PLmono to Rh under the two aggregation regimes revealed that OH• generation in tightly packed aggregates formed during slow aggregation was significantly less than the ability of aggregates formed under rapid aggregation (Figure 3b). Under the experimental conditions maintained in this study, the maximum observed decline of ROS generation was 15% and 18% for TiO2 and ZnO, respectively. Certain inorganic ions have been identified as efficient hydroxyl radical scavengers. E. coli inactivation decreased in the presence of inorganic ions, including Cl−, NO3−, S2O3−2, and SO4−2.37,38 However, the presence of organic matter in pond water was not found to impact E. coli inactivation.37 Nanoparticle coatings have also been found to impact ROS generation from TiO2 photocatalysts. Coatings of TiO2 containing SiO2 showed a significant inhibitory response toward ROS generation. In contrast, organic coatings, such as trimethylchlorosilane and poly(methyl methacrylate) exhibited little ROS suppression.39 Interestingly, sorbed humic acid on the surface of TiO2 nanoparticles was found to increase their sensitivity toward

(1)

where n is the number of primary particles in an aggregate at a given aggregate size, D is the aggregate fractal dimension, and da and dm are the average hydrodynamic diameter of the aggregate and primary particle, respectively. Therefore, the normalized PL can now be calculated as PLmono =

PL n

(2)

where PLmono is the normalized PL intensity. When comparing PLmono to the corresponding Rh it becomes clear that TiO2 aggregates have the same hydroxyl radical generation capability, regardless of the aggregation rate (Figure 3a). This strengthens 6937

dx.doi.org/10.1021/es202009h | Environ. Sci. Technol. 2012, 46, 6934−6941

Environmental Science & Technology

Article

where rm is the radius of the primary particle. A key parameter was found to be R

visible light, allowing them to degrade an organic contaminant with visible light excitation only.40 In contrast, it was observed that humic acid quenched reactive species generated from TiO2 in the presence of UV light, by as much as 52%. However, the quenching was concentration dependent, with reaction rates declining 28%, 32%, and 52% in the presence of 0.5, 5, and 20 mg/L humic acid, respectively.41 The decline in hydroxyl radical generation observed in our study is less than that reported from other causes, including the presence of large concentrations of organics and inorganic ions, but is nonetheless significant. These observations can be described by a kinetic model that considers aggregate size, structure (fractal dimension), and photon absorption to predict the impact of aggregation on OH• generation (Figure 4) in analogous form to a model we

R=

∫λ

I(λ)σ(λ)abs dλ hv(λ)

(5)

where I(λ) is the excitation light intensity at a given wavelength, hv(λ) is the energy of a photon at a given wavelength, and σ(λ)abs is the absorption cross-section at a given wavelength. The extinction coefficient of a suspended material at a given wavelength can be expressed using the Beer−Lambert law extinction(λ) =

Abs(λ) cl

(6)

where Abs(λ) is the photoabsorption measured using UV−vis spectroscopy, c is the concentration of the material, and l is the path length. The extinction cross section can be expressed as extiction(λ) = σ(λ)abs + σ(λ)scat

where σ(λ)abs is the absorption cross section and σ(λ)scat is the scattering cross section. For particles much smaller than the incident beam wavelength, σ(λ)scat can be neglected.42 Additionally, while larger particles scatter more light, Abs(λ) decreases over time as aggregate size increases (Figure S5). This indicates that if σ(λ)scat does increase significantly due to aggregation, we would expect to see an overall increase in excitation(λ). However, this was not observed (Figure S5). Thus, by following temporal changes to Abs(λ) in an aggregating sample, σ(λ)abs was paired with a given aggregate size and structure. Radiometer readings of the UV light source indicated a total intensity of 2.1 mW/cm2. The intensity of the excitation light source at each wavelength was estimated from lamp characteristics as specified by the manufacturer (Figure S6), which indicated that the excitation light was centered around 365 ± 25 nm. We assumed a normal distribution around 365 nm, with 99% of the energy falling within ±25 nm of the mean (365 nm). By fitting a normal distribution function around the mean of 365 nm, we could estimate I(λ) at any given wavelength (Table S2). Assuming steady-state, and assuming that [h+] = [e−], it is possible to solve eq 3, resulting in

Figure 4. Model schematic of reactions used in kinetic model. Equation terms over arrows represent reaction rates.

have previously presented to describe the effect of aggregation on fullerene reactivity.19 The model calculates the rate of hole (h+) creation: d[h+] = ΦR[particles] − k1[OH −][h+] − εD − 1k2[h+] dt [e−]

(3)

where Φ is the quantum yield of the material, R is the rate of optically induced transitions, [particles] is the molar concentration of primary particles of the semiconductor expressed per aggregate volume, [e−] is the molar concentration of surface electrons, [h+] is the molar concentration of surface holes, k1 and k2 are reaction rate constants (M−1 S1−), and ε is the porosity of the aggregate (eq 4). Reaction rate constant values used in this model are listed in Table S1. The first term on the right side of eq 3 is a h+ generation term. It is important to note that this term represents changes in the concentrations of h+ on the surface of the particles. The second term is a OH• generation term, and the third term is a h+ quenching term representing the case where electrons and holes on neighboring primary particles in an aggregate quench each other. ε represents an aggregate structural characteristic that depends on Rh and D, and represents the solid fraction of an aggregate 3−D ε = R hD − 3rm

(

)

(7)

[h+] =

− k1[OH −] +

k12[OH −]2 + 4εD − 1k2 φR[particles] 2εD − 1k2

(8)

Changes to the hydroxyl radical concentration can be expressed as k d[OH •] = k1[OH −][h+] − 3 [OH •]2 − εD − 1k4[OH •] dt 2 [e−] − 3D − 1k5[OH •][h+] − k6[OH •][TA] (9)

where [TA] is the concentration of the terephthalic acid. Since [TA] ≫ [OH•], we assume that the self-quenching term k3[OH•]2/2 can be neglected. The first term of the equation is a OH• generation term, the second term is a self-quenching expression, the third and fourth terms describe the interaction of a free OH• with a surface electron or hole, respectively, and the fifth term describes the interaction of OH• with the PL marker (TA). ε in the third and fourth terms accounts for the increased probability of a free OH• to interact with a monomer surface in a tightly packed aggregate, leading to increased

(4) 6938

dx.doi.org/10.1021/es202009h | Environ. Sci. Technol. 2012, 46, 6934−6941

Environmental Science & Technology

Article

slope of 1.0004 (Figure 5). Ideally the slope would be 1. The coefficient of correlation (R2) was determined to be 0.68; 95% confidence intervals calculated for the slope of the fitted line indicate that the ideal case (slope =1) falls within the area covered by the confidence intervals. Additionally, error bars surrounding the data indicate that a portion of the uncertainty expressed in the relatively low R2 value can be explained by variability in PL measurements. Changes in R as a function of aggregate size for TiO2 and ZnO suspensions aggregated under different conditions can be seen in Figure 6a and b. R values decrease with increasing Rh in both TiO2 and ZnO suspensions. Furthermore, for similar aggregate sizes, it is apparent that R values are lower in the slower aggregating suspensions. As the aggregate grows, particle shadowing decreases the ability of primary particles buried within the aggregate to absorb incoming exciting photons. This leads to fewer electron−hole pairs, ultimately resulting in lower OH• generation. Figure 7 demonstrates changes in εD‑1 as a function of Rh for aggregating TiO2 and ZnO suspensions. εD‑1 decreases with

quenching in aggregates with a higher fractal dimension. Assuming steady state, we can solve the above equation to yield k1[OH −][h+] [OH •] = D − 1 ε k4[h+] + εD − 1k5[h+] + k6[TA]

(10)

Changes in the hydroxylated form of TA (TAOH) can be expressed as d[TAOH ] = k6[OH •][TA] dt

(11)

For a given salt concentration, the impact of aggregation on OH• generation can be expressed as a ratio between the concentrations of TAOH. Therefore, we can write [TAOH ]1 [OH •]1 = [TAOH ]2 [OH •]2

(12)

where subtscripts 1 and 2 denote two distinct aggregation states within a given salt concentration. Using the rate constant values in Table S1, we were able to reasonably recreate the ratios between TAOH concentrations obtained experimentally (Figure 5).

Figure 7. Changes in εD−1 as a function of aggregate size and structure for aggregating semiconducting nanoparticles.

Figure 5. Observed vs expected results of OH• generation model. Within each material and ionic strength’s data series, results are relative to those obtained in t = 0. The dashed line represents the ideal situation where the observed results match the expected results perfectly. Solid lines are the 95% confidence intervals around the fitted slope, represented by the solid line running through the data points. Error bars around the data points represent the 95% confidence intervals around the PL measurements.

increasing aggregate size when the fractal dimension does not change greatly with aggregation (all TiO2 suspensions, and ZnO 0.05 M). However, for ZnO 0.02 M aggregation, εD‑1 does not change much with aggregation. The structure of the aggregate plays a key role in the quenching of both holes and free hydroxyl radicals. When the local concentration of primary particles increases, as in a tightly packed aggregate, generated holes can recombine with electrons from adjacent particles.

Plotting the observed vs expected values of [OH•]1/[OH•]2 and fitting a linear trendline forced through the origin yields a

Figure 6. Changes in the rate of optically induced transitions as a function of hydrodynamic radius in TiO2 suspensions (a) and ZnO suspensions (b). 6939

dx.doi.org/10.1021/es202009h | Environ. Sci. Technol. 2012, 46, 6934−6941

Environmental Science & Technology

Article

(2) Jones, N.; Ray, B.; Ranjit, K. T.; Manna, A. C. Antibacterial activity of ZnO nanoparticle suspensions on a broad spectrum of microorganisms. FEMS Microbiol. Lett. 2008, 279 (1), 71−76. (3) Kaegi, R.; Ulrich, A.; Sinnet, B.; Vonbank, R.; Wichser, A.; Zuleeg, S.; Simmler, H.; Brunner, S.; Vonmont, H.; Burkhardt, M.; Boller, M. Synthetic TiO2 nanoparticle emission from exterior facades into the aquatic environment. Environ. Pollut. 2008, 156 (2), 233−239. (4) Kormann, C.; Bahnemann, D. W.; Hoffmann, M. R. Photolysis of chloroform and other organic molecules in aqueous TiO2 suspensions. Environ. Sci. Technol. 1991, 25 (3), 494−500. (5) Li, Q. L.; Mahendra, S.; Lyon, D. Y.; Brunet, L.; Liga, M. V.; Li, D.; Alvarez, P. J. J. Antimicrobial nanomaterials for water disinfection and microbial control: Potential applications and implications. Water Res. 2008, 42 (18), 4591−4602. (6) Hoffmann, M. R.; Martin, S. T.; Choi, W. Y.; Bahnemann, D. W. Environmental applictions of semiconductor photocatalysis. Chem. Rev. 1995, 95 (1), 69−96. (7) Hussain, F.; Hojjati, M.; Okamoto, M.; Gorga, R. E. Review article: Polymer-matrix nanocomposites, processing, manufacturing, and application: An overview. J. Compos. Mater. 2006, 40 (17), 1511− 1575. (8) Robichaud, C. O.; Uyar, A. E.; Darby, M. R.; Zucker, L. G.; Wiesner, M. R. Estimates of Upper Bounds and Trends in Nano-TiO2 Production As a Basis for Exposure Assessment. Environ. Sci. Technol. 2009, 43 (12), 4227−4233. (9) Cherchi, C.; Gu, A. Z., Impact of Titanium Dioxide Nanomaterials on Nitrogen Fixation Rate and Intracellular Nitrogen Storage in Anabaena variabilis. Environ. Sci. Technol. 44, (21), 8302-8307. (10) Long, T. C.; Saleh, N.; Tilton, R. D.; Lowry, G. V.; Veronesi, B. Titanium dioxide (P25) produces reactive oxygen species in immortalized brain microglia (BV2): Implications for nanoparticle neurotoxicity. Environ. Sci. Technol. 2006, 40 (14), 4346−4352. (11) Mueller, N. C.; Nowack, B. Exposure modeling of engineered nanoparticles in the environment. Environ. Sci. Technol. 2008, 42 (12), 4447−4453. (12) Domingos, R. F.; Tufenkji, N.; Wilkinson, K. J. Aggregation of Titanium Dioxide Nanoparticles: Role of a Fulvic Acid. Environ. Sci. Technol. 2009, 43 (5), 1282−1286. (13) Hotze, E. M.; Phenrat, T.; Lowry, G. V. Nanoparticle Aggregation: Challenges to Understanding Transport and Reactivity in the Environment. J. Environ. Qual. 2010, 39 (6), 1909−1924. (14) Buxton, G. V.; Greenstock, C. L.; Helman, W. P.; Ross, A. Critical review of rate constants for reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals (i ̈ OH/i ̈ O-) in aqueous solution. J. Phys. Chem. Ref. Data 1988, 17 (2), 513−886. (15) Lin, H.; Huang, C. P.; Li, W.; Ni, C.; Shah, S. I.; Tseng, Y. H. Size dependency of nanocrystalline TiO2 on its optical property and photocatalytic reactivity exemplified by 2-chlorophenol. Appl. Catal., B. 2006, 68 (1−2), 1−11. (16) Maira, A. J.; Yeung, K. L.; Lee, C. Y.; Yue, P. L.; Chan, C. K. Size effects in gas-phase photo-oxidation of trichloroethylene using nanometer-sized TiO2 catalysts. J. Catal. 2000, 192 (1), 185−196. (17) Siedl, N.; Elser, M. J.; Bernardi, J.; Diwald, O. Functional Interfaces in Pure and Blended Oxide Nanoparticle Networks: Recombination versus Separation of Photogenerated Charges. J. Phys. Chem. C 2009, 113 (36), 15792−15795. (18) Tseng, Y. H.; Lin, H. Y.; Kuo, C. S.; Li, Y. Y.; Huang, C. P. Thermostability of nano-TiO2 and its photocatalytic activity. React. Kinet. Catal. Lett. 2006, 89 (1), 63−69. (19) Hotze, E. M.; Bottero, J. Y.; Wiesner, M. R. Theoretical framework for nanoparticle reactivity as a function of aggregation state. Langmuir 2010, 26 (13), 11170−11175. (20) Gilbert, B.; Lu, G.; Kim, C. Stable cluster formation in aqueous suspensions of iron oxyhydroxide nanoparticles. J. Colloid Interface Sci. 2007, 313 (1), 152−159. (21) Schaefer, D.; Martin, J.; Wiltzius, P.; Cannell, D. Fractal geometry of colloidal aggregates. Phys. Rev. Lett. 1984, 52 (26), 2371− 2374.

Also, there is an increased probability that free OH• will recombine with either electrons or holes on particle surfaces, increasing quenching. Eliminating the structural component from the model decreased the accuracy of the model, causing an increase in the predicted amount of TAOH generated. The experimental and computational results emphasize the importance of aggregation when discussing the fate of photocatalytic nanoparticles in the environment. Clearly, aggregate size and structure will determine how reactive these materials will remain in the environment. Simply assuming that aggregation will decrease their reactivity, and potential toxicity, could significantly underestimate the potential harm these materials can cause to aquatic life. Whereas other factors, such as coatings, organic matter, and ionic species will influence the impact of ROS on target organisms, aggregate size and structure will determine their production rate from photocatalytic nanomaterials. Thus, when assessing the toxicity of these materials, it is critical to evaluate aggregate size and structure, along with a host of other environmental conditions and potential ROS sinks. Furthermore, the application of photocatalytic nanoparticles in engineered settings, either in suspension or deposited on a substrate, should consider aggregate structure. In particular, deposited photocatalytic nanoparticles used in water treatment processes should consider both the thickness and density of the deposited layer, to avoid shadowing, quenching, and mass-transport limitations.

■

ASSOCIATED CONTENT

S Supporting Information *

Changes in Rh over time for TiO2 (Figure S1a) and ZnO (Figure S1b) after the addition of NaCl; changes in EPM as function of NaCl concentrations (Figure S1c); log(I) vs log(q) over time for aggregating TiO2 and ZnO nanoparticles (Figure S2a−e); multiple scattering verification (Figure S3); impact of aggregate size on hydroxyl radical generation in TiO2 (Figure S4a) and ZnO (Figure S4b) suspensions; change in photoabsorbance over time (Figure S5a−e); excitation spectra of UV bulb (Figure S6); model parameter values (Table S1); estimated light intensity for each wavelength (Table S2). This material is available free of charge via the Internet at http:// pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: 919-660-5292; fax: 919-660-5219; e-mail: wiesner@ duke.edu.

■

ACKNOWLEDGMENTS Financial support for this work was provided by the Center for the Environmental Implications of Nanotechnology, and is greatly appreciated. The National Science Foundation and the U.S. Environmental Protection Agency jointly fund the Center for the Environmental Implications of Nanotechnology. Special thanks to Stella Marinakos of Duke University for providing ZnO TEM images.

■

REFERENCES

(1) Frank, A. J.; Kopidakis, N.; van de Lagemaat, J. Electrons in nanostructured TiO2 solar cells: transport, recombination and photovoltaic properties. Coord. Chem. Rev. 2004, 248 (13−14), 1165−1179. 6940

dx.doi.org/10.1021/es202009h | Environ. Sci. Technol. 2012, 46, 6934−6941

Environmental Science & Technology

Article

(22) Elimelech, M.; Gregory, J.; Jia, X.; Williams, R. Particle Deposition and Aggregation: Measurement, Modelling and Simulation; Butterworth-Heinemann, 1998. (23) Ohtani, B.; Bowman, R. M.; Colombo, D. P.; Kominami, H.; Noguchi, H.; Uosaki, K. Femtosecond diffuse reflectance spectroscopy of aqueous Titanium (IV) oxide suspension: Correlation of electronhole recombination kinetics with photocatalytic activity. Chem. Lett. 1998, 579−580. (24) Okuda, M.; Tsuruta, T.; Katayama, K. Lifetime and diffusion coefficient of active oxygen species generated in TiO2 sol solutions. Phys. Chem. Chem. Phys. 2009, 11 (13), 2287−2292. (25) Hirakawa, T.; Nosaka, Y. Properties of O2·− and OH· formed in TiO2 aqueous suspensions by photocatalytic reaction and the influence of H2O2 and some ions. Langmuir 2002, 18 (8), 3247−3254. (26) Ishibashi, K.; Fujishima, A.; Watanabe, T.; Hashimoto, K. Quantum yields of active oxidative species formed on TiO 2 photocatalyst. J. Photochem. Photobiol., A 2000, 134 (1−2), 139−142. (27) Saran, M.; Summer, K. H. Assaying for hydroxyl radicals: Hydroxylated terephthalate is a superior fluorescence marker than hydroxylated benzoate. Free Radical Res. 1999, 31 (5), 429−436. (28) Jiang, J. K.; Oberdorster, G.; Biswas, P. Characterization of size, surface charge, and agglomeration state of nanoparticle dispersions for toxicological studies. J. Nanopart. Res. 2009, 11 (1), 77−89. (29) Teleki, A.; Wengeler, R.; Wengeler, L.; Nirschl, H.; Pratsinis, S. E. Distinguishing between aggregates and agglomerates of flame-made TiO2 by high-pressure dispersion. Powder Technol. 2008, 181 (3), 292−300. (30) Ball, R.; Witten, T. Particle aggregation versus cluster aggregation in high dimensions. J. Stat. Phys. 1984, 36 (5), 873−879. (31) Belko, A.; Nikitin, A.; Strekalí, N.; German, A. Fractal structure of gold clusters formed under vacuum deposition on dielectric substrates. J. Surf. Invest.: X-Ray, Synchrotron Neutron Tech. 2009, 3 (3), 338−342. (32) Daneshvar, N.; Salari, D.; Khataee, A. Photocatalytic degradation of azo dye acid red 14 in water on ZnO as an alternative catalyst to TiO2. J. Photochem. Photobiol., A 2004, 162 (2−3), 317− 322. (33) Khodja, A. A.; Sehili, T.; Pilichowski, J. F.; Boule, P. Photocatalytic degradation of 2-phenylphenol on TiO2 and ZnO in aqueous suspensions. J. Photochem. Photobiol., A 2001, 141 (2−3), 231−239. (34) Guillard, C.; Puzenat, E.; Lachheb, H.; Houas, A.; Herrmann, J. M. Why inorganic salts decrease the TiO2 photocatalytic efficiency. Int. J. Photoenergy 2005, 7 (1), 1−9. (35) Mascolo, G.; Ciannarella, R.; Balest, L.; Lopez, A. Effectiveness of UV-based advanced oxidation processes for the remediation of hydrocarbon pollution in the groundwater: A laboratory investigation. J. Hazard. Mater. 2008, 152 (3), 1138−1145. (36) Sahoo, C.; Gupta, A.; Pal, A. Photocatalytic degradation of Methyl Red dye in aqueous solutions under UV irradiation using Ag+ doped TiO2. Desalination 2005, 181 (1−3), 91−100. (37) Ireland, J. C.; Klostermann, P.; Rice, E. W.; Clark, R. M. Inactivation of Escherichia coli by titanium dioxide photocatalytic oxidation. Appl. Environ. Microbiol. 1993, 59 (5), 1668−1670. (38) Rincon, A.-G.; Pulgarin, C. Effect of pH, inorganic ions, organic matter and H2O2 on E. coli K12 photocatalytic inactivation by TiO2: Implications in solar water disinfection. Appl. Catal., B 2004, 51 (4), 283−302. (39) Carlotti, M. E.; Ugazio, E.; Sapino, S.; Fenoglio, I.; Greco, G.; Fubini, B. Role of particle coating in controlling skin damage photoinduced by titania nanoparticles. Free Radical Res. 2009, 43 (3), 312−322. (40) Schmelling, D. C.; Gray, K. A.; Kamat, P. V. The influence of solution matrix on the photocatalytic degradation of TNT in TiO2 slurries. Water Res. 1997, 31 (6), 1439−1447. (41) Wang, Y.; Hong, C. S.; Fang, F. Effect of solution matrix on TiO2 photocatalytic degradation of 2-chlorobiphenyl. Environ. Eng. Sci. 1999, 16 (6), 433−440.

(42) Leatherdale, C.; Woo, W. K.; Mikulec, F.; Bawendi, M. On the absorption cross section of CdSe nanocrystal quantum dots. J. Phys. Chem. B 2002, 106 (31), 7619−7622. (43) Seyed-Dorraji, M.; Daneshvar, N.; Aber, S. Influence of inorganic oxidants and metal ions on photocatalytic activity of prepared zinc oxide nanocrystals. Global NEST J. 2009, 11 (4), 535− 545.

6941

dx.doi.org/10.1021/es202009h | Environ. Sci. Technol. 2012, 46, 6934−6941